The Problem with Using Zip Codes or a Radius to Pull Direct Mail Lists
In my post entitled “Not All Market Areas are Created Equally: Part 1” I discussed a couple of different ways that MAR differs of other research firms generating Market Area Reports. In this post, I want to look at other, more typical ways that marketing professionals tend to view the area that they service – zip codes and a radius. I also want to look at how using a zip code or a radius to pull a direct mail list can result in a lower than ideal response rate.
I hear it a lot, a customer wants to pull a list for targeting new prospects. We go through the discussion of who and where to target, and 9 times out of 10, the client is confident about their response on who they want to target (which demographics), but less certain when we talk about which areas of the community they’d like to hit. They generally think of a radius or a set of zip codes, while less often do they think about targeting neighborhoods. When we get in to the neighborhood conversation, the client tends to think more about doing EDDM (Every Door Direct Mail), which does not allow the user to filter out addresses they’re not interested in targeting. (For more information on EDDM vs Direct Mail targeting, see my Journal post). So let’s look at the other most common ways a marketer views their market area – zip codes analysis and radius analysis.
As we’re all well aware, zip codes are geographic areas defined by the USPS. They are universally used for grouping addresses in to areas, and they are usually one of the first things that a marketer thinks of when they want to identify an area to target. The issue that MAR has with zip code analysis are that they’re usually pretty big as far as area. What MAR likes to use as an alternative unit of analysis are block groups. Block groups are a geographic area defined by the census that are much smaller than zip codes. They usually follow streets pretty nicely, and we tend to think of them more as neighborhoods. See the image to the left to get an idea of the difference in size between zip codes and block groups.
So whats the problem with analyzing zip codes? Well as I mentioned, their size. Lets look at Los Angeles County. In the county, Google tells me that there are about 500 zip codes in Los Angeles county. Sounds like a lot right? Well if we intersect the Los Angeles county boundary with our block group boundary file, we’ll see that there about 11,000 block groups! Over 20 times as many. A quick spatial function tells me that in LA, these zip codes have close to an average area about 8 square miles, where the block groups have an average area of about 1.5 square miles. Statistics 101 tells us that its difficult to de-aggregate data without making some assumptions and that its always better to have as granular level data as possible, when doing an analysis. This is the primary reason MAR uses block groups.
Another reason that MAR prefers block groups is that the census makes available block groups demographics. Hundreds of demographic variables aggregated at the block group level, and its smallest reliable source of block group demographic data out there. That means that when we’re performing our analysis, we can use things like average income of the block, median home value, distance of the block group, percentage of families, and other variables to compare to our business and use them for targeting. This is where the Direct Mail Analytics piece of the Market Area Report comes in.
Several pages in the Market Area report that MAR delivers to its clients are dedicated to what we call our Direct Mail Analytics section. Its a map that establishes what we call “block group attractiveness”. MAR defines block group attractiveness as how receptive a block group is going to be to our marketing efforts based on some of the variables listed above, and one other key variable – block group penetration. Block group penetration is simply defined as the number of clients living in a block group divided by the total number of households in the block. The penetration, along with some of the other block group demographic variables, help us to establish its attractiveness. MAR assigns an attractiveness score, then color codes the block groups based on this score – green means high attractiveness, yellow means medium, red indicates low. See the map to the right.
One can easily see that when we use MAR’s model for establishing attractiveness, then over lay our zip code map, there tends to be a huge discrepancy. Let me try and simplify MAR’s attractiveness model down some. Take for instance just the variables of penetration and distance. This is how far away the block is and how many current clients live there. We’ll make a simple coloring matrix of green if we have high penetration and the block is close to our business, red if penetration is low and the distance is far away, and everything else will be yellow. Look at the Westchase zip code of 33626. A zip code analysis WILL tell us that the 33626 zip is one that we should target. But if we look at the block groups, we’ll see that we’ll pull in some names of prospect households way up on the north side of the zip. And if we target 33625 which is also close to the business, we’ll pull in names on the other side of the expressway, which will almost definitely generate zero response.
Lets briefly look at the Radius analysis where the principal is effectively the same. If we look at the map on the left, we’ll see that after we’ve established our block group attractiveness, there isn’t a radii ring that we can draw that will not pull in households that do not have a low likelihood of response. Once again, many marketers think that they’ll draw a circle around their business, and run to count the demographic profile they’re interested in. What they fail to realize is that some of the neighborhoods that they’re going to pull in, most likely have a poor geographic relationship to the business. Observe on the radius map how the attractiveness changes seem to be correlated to network of roads. Blocks on the east of the road 589 and south of 580 have lower attractiveness compared to neighborhoods that are right on the other side of the street. This is a very common phenomenon – people generally don’t like to hop on the interstate when the business is not a destination location such a mall or a unique restaurant.
This concludes my thoughts on the problem with using zip codes or a radius to pull direct mail lists. I hope this Journal entry has helped you to have more clarity on why radius and zip code analyses are less than optimal ways for determining your market area and where to target. Drop us a line if you would like to discuss these ideas further in more detail.